stator insulation quality assessment for high 6701_chap01

ADVANCED NONDESTRUCTIVE EVALUATION IIIn Two Volumes, with CD-ROM - Proceedings of the International Conference on ANDE 2007© World Scientific Publishing Co. Pte. Ltd.http://www.worldscibooks.com/engineering/6701.html

1

STATOR INSULATION QUALITY ASSESSMENT FOR HIGH

VOLTAGE MOTORS BASED ON PROBABILITY

DISTRIBUTIONS

HEE-DONG KIM, CHUNG-HYO KIM

KEPCO Korea Electric Power Research Institute, 103-16 Munji-dong, Yuseong-gu,

Daejeon 305-380, Korea

Stator insulation quality assessment for high voltage motors is a major issue for reliable

maintenance of industrial and power plants. To assess the condition of stator insulation,

nondestructive tests are performed on the sixty coil groups of twelve motors. The stator

winding of each motor is classified into five coil groups; one group with healthy

insulation and four groups with four different types of artificial defects. To analyze the

breakdown voltage statistically, Weibull distribution is employed for the tests on the fifty

coil groups of ten motors. The 50th percentile values of the measured breakdown

voltages based on the statistical data of the five coil groups of ten motors were 26.1kV,

25.0kV, 24.4kV, 26.7kV and 30.5kV, respectively. Almost all of the failures were

located in the line-end coil at the exit of the core slot. The breakdown voltages and the

types of defects are strongly related to the stator insulation tests such as the dissipation

factor and ac current. It is shown that the condition of the motor insulation can be

determined from the relationship between the probability of failure and the type of defect.

1. Introduction

Industrial surveys and other studies on machine reliability show that the winding

insulation is the most vulnerable component in high voltage (HV) rotating and

stationary electric machines [1,2]. Since the insulation of the electric machine

windings is continuously exposed to a combination of thermal, electrical,

mechanical, and environmental stresses during operation, the insulation material

deteriorates gradually over time. Thermal stress causes formation of voids in the

mica barrier, weak bonding between layers, and delamination. Electrical and

mechanical stresses also contribute to formation of voids in the insulation, and

environmental stress causes rupture of the chemical bonds. The synergistic effect

of the stresses causes gradual deterioration of stator winding insulation, which

leads to eventual electrical failure.

There are many different types of nondestructive and destructive tests used

for evaluating the degree of insulation degradation in HV motors.

Nondestructive tests include measurements of insulation resistance, ac current,

capacitance, dissipation factor (tanδ), and partial discharge (PD). To assess the


2

insulation condition in power plants, periodic nondestructive tests must be

performed in HV motors [3,4]. Destructive testing of the stator winding can be

conducted by gradually increasing ac voltage until the insulation fails. Analysis

of breakdown voltage is considered a more sensitive indicator and a more

reliable way of estimating insulation deterioration than nondestructive tests.

This paper describes several test results from both nondestructive and

destructive tests in sixty coil groups of twelve HV motors. Weibull probability

distribution is employed to analyze the ac breakdown voltage from a selected

group of coils.

2. Test Specimen and Experimental Procedure

To assess the deterioration of stator winding insulation in HV motors, a total of

576 coils were manufactured for twelve motors (48 slots), where the motor is a

6.6kV motor. All 48 coils for each motor (12 turn coil) are inserted in the stator

slot when performing the tests. The 48 coils are classified into five groups

where one group of coils (E) is healthy, artificial defects are introduced into four

coil groups (A~D). The four different types of coils inserted in the four coils

groups are shorted turn (strand), internal separation between conductor and

groundwall (GW) insulation, large void within the GW insulation, and removal

of semi-conductive tape. All the coil groups other than coil group D, have the

semi-conductive tape on the outer surface of the slot section. Table 1

summarizes the description of each coil group (A~E), and the number of

windings with each defect. A photograph of the test motor with the five coil

groups of the HV motor is shown in Figure 1. Twelve motors were manufactured

with identical coil groups with the artificial defects. Nondestructive tests were

performed on the individual coil groups of twelve motors to obtain the statistical

data.

The ac current and dissipation factor (tanδ) were measured using a

commercial Schering bridge in all five coil groups of motor stator windings. The

Schering bridge consists of a HV power supply (Tettex, Type 5283), and bridge

(Tettex, Type 2818). The ac current and tanδ measurements were obtained

between 0.95kV and 6.6kV for all five coil groups. ∆I is the difference between

the actual measurement of the current at 6.6kV and ideal expected current at

6.6kV estimated based on the increase in current. ∆tanδ is the difference

between the tanδ measurements at 2kV and at 6.6kV. PD measurements were

also obtained in accordance with IEC 270 using a wide band (40~400 kHz)

partial discharge detection system (TE-571) in a calibrated measuring circuit.

The AC breakdown test was performed using a variable ac power supply.


3

Table 1. Defect classifications in five coil groups of motor

Coil Group Defect Classifications Winding Numbers

A Shorted turn of strand 10

B Internal separation between conductor

and insulation 10

C Large void within insulation 8

D Removal of semi-conductive tape 10

E Healthy 10

Figure 1. Five coil groups of motor with healthy and artificial defects

3. Results and discussion

Before the ac breakdown test was performed, nondestructive tests that include

the ac current, dissipation factor (tanδ), and partial discharge tests were

performed on the sixty coil groups of twelve motors. The measurements of the ac

current and tanδ were obtained with the voltage between 0.95kV and 6.6kV for

each individual coil group for the twelve motors. For each coil group (A~E) of

the twelve motors, the minimum and maximum values of ∆I and ∆tanδ were

removed in the statistical analysis (ten of sixty coil groups for the twelve motors

have been removed). The average of the ∆I and ∆tanδ measurements with the

minimum and maximum values removed are summarized in Table 2.

According to [5], coils with the values of ∆I and ∆tanδ lower than 8.5% and

6.5%, respectively, are acceptable at 6.6kV. It can be seen in Table 2 that the

measurements of ∆I and ∆tanδ for coil groups A and E have the lowest values.

The values of ∆I and ∆tanδ being below 8.5% and 6.5% indicate that the stator

winding insulation in coil groups A and E is in serviceable condition. The values

of the ∆I and ∆tanδ measurements in coil groups B, C, and D are above the 8.5%


4

and 6.5% threshold, which indicate that they must be replaced since they have

deteriorated significantly. Coil group B appears to be prone to delamination and

debonding between the conductor and insulation. The tanδ values of coil group

C at 0.95kV and at 6.6kV were measured to be 6.2~8.0% and 11.0~16.0%,

respectively. The tanδ values of coil group D (no semiconductive tape) at

0.95kV and 6.6kV were between 0.5~2.0% and 11.0~22.4%, respectively, which

was roughly eight times higher than those for coil group E (healthy).

Table 2. The measured results of ∆I and ∆tanδ

Coil Group △tanδ [%] △I [%]

A 4.31 5.35

B 6.62 10.90

C 7.74 9.87

D 11.03 15.21

E 1.37 1.89

Table 3. The results of PD measurement

PD magnitude [pC]

Coil Group Noise

[pC]

DIV

[kV]

3.81

[kV]

4.76

[kV] 6 [kV] 6.6 [kV]

A 410 4.3 1100 2200 3800 5600

B 470 3.4 1700 3700 6000 6900

C 430 3.7 1300 2500 6300 8800

D 450 3.3 2900 8700 14000 17000

E 420 3.5 1300 2000 4000 5700

As in the case of the ∆I and ∆tanδ measurements, the data with the minimum

and maximum values of PD magnitudes were removed in the statistical analysis

for each coil group (A~E). For each coil group, the external noise, discharge

inception voltage (DIV) and the average PD magnitude with the minimum and

maximum values removed are summarized in Table 3. The PD magnitudes were

measured for each individual coil group at 3.81kV, 4.76kV, 6kV, and 6.6kV,

respectively. As the voltage was increased from 3.81kV to 6.6kV, the PD

magnitudes increased, as expected. It can be seen that the PD magnitudes at line-

to-ground voltage (3.81kV) and at 4.76kV are in good condition for the five coil

groups. The PD magnitudes at 6.6kV were 5,600pC, 6,900pC, 8,800pC, and

5,700pC for coil groups A, B, C, and E, and unacceptably high for coil D at the

same voltage. The PD magnitudes in coil group D are high enough to cause

significant damage to the insulation. The PD pattern in the coil group A, C, and


5

E measurements indicated internal discharges, and the PD patterns of coil groups

B and D indicated discharge at conductor surface and slot discharges,

respectively, as expected from the artificial defects. Although the PD magnitude

of coil group D are much higher than that of coil group C, internal discharge

causes more serious insulation problems than slot discharge.

The AC breakdown test, which is a test applied to HV motor insulation to

test suitability of service, was performed on the sixty coil groups of twelve

motors to confirm the results of the nondestructive test. As in the nondestructive

tests, the minimum and maximum breakdown voltages were removed from the

measurements of each coil group (A~E). The results of the average of the

breakdown voltage measurements from the ten coils groups are summarized in

Table 4. Breakdown of coil groups A, B, C, D and E occurred at an average of

25.9kV, 24.8kV, 24.1kV, 26.5kV and 30.4kV, respectively, which indicates that

the stator insulation is in serviceable condition. The breakdown voltage of the

healthy coil group (E) is higher than that of the coil groups with artificial defects

(A~D). The lowest breakdown voltage was observed in the coil group C with a

large void in the bulk of the GW insulation. Most of the failures occurred in the

line-end coil at the exit of the core slot section; only the failures in two coils

occurred in slot section.

Table 4. The results of the average breakdown voltage

Coil Group Breakdown Voltage [kV] Failure Location

A 25.9 Line-end coil

B 24.8 Line-end coil

C 24.1 Line-end coil

D 26.5 Line-end coil

E 30.4 Line-end coil

Figure 2 shows the Weibull plot of the ac breakdown voltage for the five coil

groups A~E described in Table 1. It is possible to estimate the failure rate of

fifty coil groups of ten motors from the Weibull distribution analysis for

breakdown voltage. The 50th percentile (median) values of the measured

breakdown voltages based on statistical data in coil groups A, B, C, D and E

were 26.1kV, 25.0kV, 24.4kV, 26.7kV and 30.5kV, respectively. It can be seen

in Table 4 that data is similar to the average values of the measured breakdown

voltage. This results shows that destructive testing is more reliable although

nondestructive testing is performed to estimate insulation deterioration rate for

breakdown voltage.

Figure 2. Weibull plot of breakdown voltage for five coil groups


6

4. Conclusion

The measurements of ∆I and ∆tanδ indicated that the insulation of coil groups A

and E are n serviceable condition, but the insulation of coil groups B, C, and D

are in poor condition. The PD magnitudes at 4.76 kV were around 2200pC,

3700pC, 2500pC, 8700pC, and 2000pC for coil groups A, B, C, D and E,

respectively. These results indicated that that the stator insulation of the five coil

groups is in good serviceable condition. It has been observed in the tests that

internal discharge (coil group C) causes more serious insulation problems than

slot discharge (coil group D). The 50th percentile (median) values of measured

breakdown voltages based on statistical data in coil groups A, B, C, D and E

were 26.1kV, 25.0kV, 24.4kV, 26.7kV, and 30.5kV, respectively. This data is in

close proximity to the average of the breakdown voltage measurements. It has

also been observed that almost all of the failures were located in the line-end coil

at the exit from the core slot section. The test results show that destructive

testing is more reliable although nondestructive testing is performed to estimate

insulation deterioration rate for breakdown voltage.

References

1. H.A. Toliyat and G.B. Kliman, Handbook of Electric Motors, (Marcel

Dekker, 2004).

2. H.G. Sedding, R. Schwabe, D. Levin, J. Stein and B.K. Gupta, IEEE

EIC/EMCW Conf, 455 (2003).

3. B.K. Gupta and I.M. Culbert, IEEE Trans. on EC, Vol. 7, 500 (1992).

4. G.C. Stone, H.G. Sedding, B.A. Lloyd and B.K. Gupta, IEEE Trans. on EC,

Vol. 3, 833 (1988).

5. H. Yoshida and U. Umemoto, IEEE Trans. on EI, Vol. 6, 1021 (1986).

stator insulation quality assessment for high 6701_chap01

Documents